Technetium-99m sestamibi (MIBI) is thought to be passively taken up by metabolically active tumour cells and effluxed from them by P-glycoprotein (Pgp). This 170-kDa membrane-bound protein, encoded by the MDR-1 gene, acts as an energy-dependent efflux pump for several antineoplastic agents, resulting in multidrug resistance. For this reason, it is of interest whether the tumour's response to chemotherapy can be predicted by MIBI single-photon emission tomography (SPET). In this study, MIBI SPET was compared with thallium-201 (Tl) SPET using magnetic resonance imaging as a guide in 16 patients with untreated brain tumours [ten glioblastomas (GBs), two anaplastic astrocytomas (AAs), two low-grade gliomas (LGASs) and two metastatic brain tumours) and in four patients who had received treatment for with brain tumours (two GBs, two AAs). In addition, we investigated the expression of the MDR-1 gene and its product Pgp in the same patients, and compared the results with MIBI SPET findings. MIBI, as well as Tl, was highly accumulated and retained in the enhanced region of malignant gliomas. In addition, MIBI SPET yielded sharp and well-contrasted images, and the margin of the tumour was more clearly defined than with Tl SPET due to a good signal-to-noise ratio. Follow-up MIBI SPET in patients who had received therapy showed marked uptake in a patient with malignant transformation, who deteriorated clinically. Patients with no uptake on MIBI SPET showed no sign of recurrence. Semiquantitative analysis of untreated patients showed a relationship between the early uptake index (UI, ratio of average count/pixel in the lesion to that in the contralateral area on early images) and the degree of malignancy (early UI = 1.08+/-0.06 in LGASs, 4.10+/-0.84 in AAs, 5.71+/-3.47 in GBs, and 7.52+/-1.52 in metastatic brain tumours). The retention index (RI, ratio of delayed to early UI) of MIBI was significantly lower than that of Tl in metastatic brain tumours (P<0.05), but not in malignant gliomas. Histological and biological investigation of gliomas showed that the MDR-1 gene and its product Pgp were expressed only in normal endothelial cells and not in tumour cells or proliferating endothelial cells; Pgp tended to decrease as the degree of malignancy rose. Hence, the presence of Pgp and the grade of malignancy were inversely related in gliomas. By contrast, immunohistochemical study showed strong accumulation of Pgp in metastatic brain tumour cells. These histopathological findings and MIBI SPET findings are compatible with experimental data; MIBI was washed out by Pgp. The main cause of chemoresistance is probably not an increasing drug efflux by Pgp in gliomas. Thus, MIBI SPET is useful for detecting the active lesions, but may not be useful for predicting the response to chemotherapy in gliomas.